Sistema de visão multispectral para plataforma robótica de inspeção termográfica em subestação de energia

Abstract

Performance monitoring of production equipment is essential for maintenance planning of any productive organization. Electric Power production and distribution companies equipments are subject to electromechanical requests causing thermal anomalies indicating possible failure trends that compromise their operation. Due to the public nature of the service provided by these companies, unscheduled outages for maintenance are discouraged. In this context, the Thermography, a non-contact temperature measurement technique, presents itself as a practical and reliable tool to electrical systems monitoring becoming, therefore, an alternative to the construction of automated inspection systems. However, it is clear that the task of identifying the inspected equipment, parameterization of the acquisition, segmentation and extraction of the thermal signal is complex and rely heavily on the application to which they are directed. This work presents a mult-spectral 3D vision system for the autonomous thermographic inspection applied to Electric Substations in different lighting conditions, distance, and angle of view dealing also with line of sight obstructions to the inspected equipment. This system is able to identify the equipment under inspection, adjusts the measuring parameters, capture and extracts the thermal signal from the scene based on disparity maps. As a test, this system has been embedded on a mobile platform inspection and the results show that the proposal is a practical and viable alternative to automated thermographic systems